Diagnostic process analysis system

Abstract

Upon evaluation of a value of a diagnostic process, the value of a diagnostic process is evaluated not based on a simple cost but on a cost required for all processes of a patient who was in the diagnostic process through a follow-up survey. Diagnostic processes that are not relevant to a target diagnostic process are eliminated, clustering is performed on patients to divide the patients into clinically meaningful homogeneous groups, and the target diagnostic process is evaluated for each of the homogeneous groups. For the purpose, importance scores of data pieces of the clinical data are calculated and the relevant data is output using the output result of the medical knowledge extraction unit, clustering is performed on patients in the clinical data, and a clinical index and a cost are output for each of the clusters.

The invention claimed is: 
     
       1. A diagnostic process analysis system that analyzes cost-effectiveness of a diagnostic process by using a database storing clinical data including patient information, medical concept information indicating medical concepts, and text data, the system comprising a computer programmed to:
 accept input of a first diagnostic process,
 extract, from the text data, relevance information indicating relevance between different medical concepts regarding the medical concept information of respective data pieces of the clinical data that are previously defined; 
 calculate importance scores of the data pieces of the clinical data by using the relevance information indicating the relevance between different medical concepts regarding the medical concept information of respective data pieces of the clinical data that are previously defined; 
 extract a second diagnostic process having a relevance to the first diagnostic process, which has been accepted, higher than a predetermined threshold based on the importance scores; 
 perform clustering on the patient information of the clinical data based on the second diagnostic process and the calculated importance scores to generate patient groups; and 
 calculate a clinical index and a cost of the second diagnostic process for each of the patient groups; 
 
 wherein the text is medical literatures; 
 wherein extracting the relevance information includes extracting evidence levels from study levels indicated by the medical literatures; 
 wherein calculating the importance scores includes calculating the importance scores of the data pieces of the clinical data based on the evidence levels and co-occurrence degrees between the medical concepts obtained by extracting the relevance information; 
 wherein a co-occurrence degree between a plurality of medical concepts is a number of medical literatures that each include all of the plurality of medical concepts; 
 wherein diagnostic processes that do not have relevance higher than the predetermined threshold based on the importance scores are eliminated from further consideration; and 
 wherein one or more second diagnostic processes having relevance higher than the predetermined threshold are further considered by using the calculated clinical indices and costs of the one or more second diagnostic processes for the patient groups, combinations of diagnostic processes are evaluated, and a combination of diagnostic processes to be improved is identified. 
 
     
     
       2. The diagnostic process analysis system according to  claim 1 , wherein
 calculating the importance scores includes evaluating deterioration degrees of the clinical index and calculating the importance scores further based on the evaluated deterioration degrees of the clinical index, the deterioration degrees classifying deterioration of the clinical data. 
 
     
     
       3. The diagnostic process analysis system according to  claim 1 , wherein the co-occurrence degree between the plurality of medical concepts is the number of medical literatures that each include all of a plurality of medical terms representing the plurality of medical concepts. 
     
     
       4. The diagnostic process analysis system according to  claim 2 , wherein
 calculating the importance scores includes, for each of the diagnostic processes, extracting a group of patients who have been in the diagnostic process and a group of patients who have not been in the diagnostic process from the database, calculating a difference between the clinical indices of the two extracted groups, and evaluating the deterioration degree of the clinical index based on the calculated difference of the clinical indices. 
 
     
     
       5. The diagnostic process analysis system according to  claim 1 , wherein
 performing clustering on the patient information of the clinical data includes accepting input of the number of clusters upon performing the clustering, and performing clustering on patients of the clinical data based on the accepted number of clusters. 
 
     
     
       6. The diagnostic process analysis system according to  claim 4 , wherein
 the deterioration degree of the clinical index is evaluated based on the difference between the clinical indices of the two extracted groups, and the importance score is calculated based in part on the difference between the clinical indices of the two extracted groups. 
 
     
     
       7. The diagnostic process analysis system according to  claim 4 , wherein
 calculating the importance scores includes, for each one of the diagnostic processes, calculating the importance score based on the evidence levels and co-occurrence degrees between the medical concepts obtained by extracting the relevance information for the data pieces of the clinical data that are relevant to said one of the diagnostic processes, and the evaluated deterioration degree of the clinical index. 
 
     
     
       8. The diagnostic process analysis system according to  claim 7 , wherein
 the deterioration degree of the clinical index is evaluated based on the difference between the clinical indices of the two extracted groups, and the importance score is calculated based in part on the difference between the clinical indices of the two extracted groups. 
 
     
     
       9. The diagnostic process analysis system according to  claim 8 , wherein
 the clinical index includes information of length of stay in hospital and hospital readmission. 
 
     
     
       10. The diagnostic process analysis system according to  claim 7 , wherein
 the clinical index includes information of length of stay in hospital and hospital readmission. 
 
     
     
       11. The diagnostic process analysis system according to  claim 6 , wherein
 the clinical index includes information of length of stay in hospital and hospital readmission. 
 
     
     
       12. The diagnostic process analysis system according to  claim 4 , wherein
 the clinical index includes information of length of stay in hospital and hospital readmission. 
 
     
     
       13. The diagnostic process analysis system according to  claim 2 , wherein
 the clinical index includes information of length of stay in hospital and hospital readmission. 
 
     
     
       14. The diagnostic process analysis system according to  claim 1 , wherein
 the clinical index includes information of length of stay in hospital and hospital readmission. 
 
     
     
       15. The diagnostic process analysis system according to  claim 3 , wherein the plurality of medical terms representing the plurality of medical concepts are extracted from abstracts of the medical literatures. 
     
     
       16. The diagnostic process analysis system according to  claim 1 ,
 wherein the identified combination of diagnostic processes are improved. 
 
     
     
       17. A diagnostic process analysis method for analyzing cost-effectiveness of a diagnostic process by using a database storing clinical data including patient information, medical concept information indicating medical concepts, and text data, the method comprising:
 accepting input of a first diagnostic process, 
 extracting, from the text data, relevance information indicating relevance between different medical concepts regarding the medical concept information of respective data pieces of the clinical data that are previously defined; 
 calculating importance scores of the data pieces of the clinical data by using the relevance information indicating the relevance between different medical concepts regarding the medical concept information of respective data pieces of the clinical data that are previously defined; 
 extracting a second diagnostic process having a relevance to the first diagnostic process, which has been accepted, higher than a predetermined threshold based on the importance scores; 
 performing clustering on the patient information of the clinical data based on the second diagnostic process and the calculated importance scores to generate patient groups; 
 calculating a clinical index and a cost of the second diagnostic process for each of the patient groups; 
 wherein the text is medical literatures; 
 wherein the extracting includes extracting evidence levels from study levels indicated by the medical literatures; 
 wherein the importance scores of the data pieces of the clinical data are calculated based on the evidence levels and co-occurrence degrees between the medical concepts obtained by extracting the relevance information; and 
 wherein a co-occurrence degree between a plurality of medical concepts is a number of medical literatures that each include all of the plurality of medical concepts; 
 eliminating from further consideration diagnostic processes that do not have relevance higher than the predetermined threshold based on the importance scores are eliminated from further consideration; and 
 further considering one or more second diagnostic processes having relevance higher than the predetermined threshold by using the calculated clinical indices and costs of the one or more second diagnostic processes for the patient groups, evaluating combinations of diagnostic processes, and identifying a combination of diagnostic processes to be improved. 
 
     
     
       18. The diagnostic process analysis method according to  claim 17 , wherein the co-occurrence degree between the plurality of medical concepts is the number of medical literatures that each include all of a plurality of medical terms representing the plurality of medical concepts. 
     
     
       19. The diagnostic process analysis method according to  claim 18 , wherein the plurality of medical terms representing the plurality of medical concepts are extracted from abstracts of the medical literatures. 
     
     
       20. The diagnostic process analysis method according to  claim 17 , further comprising:
 improving the identified combination of diagnostic processes.